overcoming transmission constraints: energy storage … · overcoming transmission constraints:...

34
Overcoming Transmission Constraints: Energy Storage and Wyoming Wind Power Mindi Farber-DeAnda and Delma Bratvold Science Applications International Corporation Tim Hennessy VRB Power Systems Dale Hoffman and Thomas Fuller Wyoming Business Council Under a State Energy Grant from the U.S. Department of Energy, Energy Storage Systems Program June 2007

Upload: doankien

Post on 05-Jun-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

Overcoming Transmission Constraints:Energy Storage and

Wyoming Wind Power

Mindi Farber-DeAnda and Delma BratvoldScience Applications International Corporation

Tim HennessyVRB Power Systems

Dale Hoffman and Thomas FullerWyoming Business Council

Under a State Energy Grant from the U.S. Department of Energy,Energy Storage Systems Program

June 2007

ii

Acknowledgement

This project was funded through a State Energy Program Grant from the Energy Storage SystemsProgram of the U.S. Department of Energy. Project collaborators include Dale Hoffman and ThomasFuller of the Wyoming Business Council; Mindi Farber-DeAnda, Delma Bratvold, and Victor Gorokhovof Science Applications International Corporation, Tim Hennessy of VRB Power Systems, Mark Kuntz,formerly of VRB, Craig Quist of PacifiCorp, and Brad Williams and Hans Isern, formerly of PacifiCorp.The collaborators would like to thank Dr. Imre Gyuk of the U.S. Department of Energy and John Boyesof Sandia National Laboratories for their support and funding of this grant. We would like toacknowledge the assistance of PacifiCorp, SeaWest, and VRB, in obtaining the data on wind farm andbattery operations as well as transmission congestion.

iii

Table of Contents

1 Introduction ..................................................................................................................................... 11.1 Background ............................................................................................................................ 11.2 The Site .................................................................................................................................. 11.3 Power Transmission and Congestion....................................................................................... 21.4 Flow Battery Contributions ..................................................................................................... 31.5 Project Objectives................................................................................................................... 3

2 Methods........................................................................................................................................... 42.1 Datasets .................................................................................................................................. 42.2 Tariff Rates and Rebates ......................................................................................................... 42.3 Excel Model for Storage System Analysis............................................................................... 62.4 Excel Model for Light Wind Capture ...................................................................................... 7

3 Time of Day Patterns in Data ........................................................................................................... 93.1 Transmission Line Congestion ................................................................................................ 93.2 Wind Speed and Power Generation ....................................................................................... 103.3 Battery Cycling, Transmission Congestion, and Tariffs ......................................................... 12

4 Flow Battery Assessment ............................................................................................................... 144.1 Battery Discharge and Revenue ............................................................................................ 144.2 Cash Inflow and Outflow...................................................................................................... 154.3 Effects of Curtailment........................................................................................................... 164.4 Battery Size and Energy Rate Differentials ........................................................................... 17

5 Capture of Light Winds ................................................................................................................. 195.1 Output to Grid and Revenue.................................................................................................. 195.2 Cash Inflow and Outflow Comparisons ................................................................................. 205.3 Effects of Curtailment........................................................................................................... 20

6 Conclusions ................................................................................................................................... 22Appendix A. Transmission Line Constraints ......................................................................................... 23Appendix B. Wind Speed and Power Generation .................................................................................. 25Appendix C. Model Snapshots.............................................................................................................. 28

iv

Table of Figures

Figure 1. TOT4 Transmission Congestion Profile ................................................................................... 9Figure 2. Power Generation and Wind Speed Percentiles ....................................................................... 10Figure 3. Monthly Average Wind Speed and Power Generation ............................................................ 10Figure 4. Seasonality and Duration of No-Generation Events ................................................................ 11Figure 5. No-Generation Events by Season, 2003 ................................................................................. 11Figure 6. Wind Speed and No-Generation Events by TOU and Season, 2003 ........................................ 12Figure 7. Discharge and Revenue Modeled for a 12-MWh Battery........................................................ 14Figure 8. Modeled Revenue With and Without Curtailment for a 12-MWh Battery ............................... 14Figure 9. Cash Inflow and Outflow for a 12-MWh Battery Under Three Revenue Scenarios ................. 16Figure 10. Energy Revenue and O&M Costs by Operation and Curtailment .......................................... 17Figure 11. Effect of Energy Rate Differential on Rate of Return............................................................ 18Figure 12. Energy Revenue and Output to Grid Increases from Spilled Wind ........................................ 19Figure 13. Capacity Revenue Increase and Fewer Penalties from Spilled Wind ..................................... 19Figure 14. Cash Inflow and Outflow for Spilled Wind Capture Under Three Revenue Scenarios ........... 20Figure 15. Effect on Curtailment on Spilled Wind Energy Revenue ...................................................... 21Figure A-1. Transmission Paths Impacting the Northwest ..................................................................... 23Figure A-2. Calculating TOT4 Transmission Line Loads ...................................................................... 23Figure A-3. TOT4 Daily System Load – Week in July 2003 ................................................................. 24Figure A-4. TOT4 Daily System Load – Week in December 2003 ........................................................ 24Figure A-5. Calculating TOT4 System Load Duration .......................................................................... 24Figure B-1. Foote Creek Rim I Layout................................................................................................... 25Figure B-2. Errors in Wyoming Wind Speed Measurements .................................................................. 26Figure B-3. Wind Speed and Turbine Output in Winter.......................................................................... 26Figure B-4. Wind Speed and Turbine Output in Summer ....................................................................... 27Figure C-1. Wind Battery Economic Model Input Page......................................................................... 28Figure C-2. Wind Battery Economic Model Output Page ...................................................................... 29Figure C-3. Wind Battery Economic Model Data and Calculation Page ................................................ 30

Table of Tables

Table 1. Seasonal TOU Periods Defined in Selected Tariff ..................................................................... 5Table 2. Seasonal TOU Energy and Capacity Payments in the Proxy Tariff ............................................ 5Table 3. Energy Storage System Economic Model Input Variables ......................................................... 6Table 4. Energy Storage System Economic Model Output and Projection Periods ................................... 7Table 5. Light Wind Model Input Variables ............................................................................................ 8Table 6. Transmission Line Congestion by Season and Tariff Period .................................................... 10

1

1 Introduction

This study represents a collaborative effort to investigate how energy storage can improve the economicsof a wind farm constrained by transmission line congestion. A number of tools were built to analyze theproprietary data and the findings confirm the advantage of installing energy storage to overcometransmission constraints.

1.1 Background

The windy and remote conditions of Wyoming have helped to make it home to most of the major windfarm developments in the U.S. in recent years. Wyoming is a significant exporter of electricity toWestern states because power generation from the various local resources, including coal, natural gas, andwind, far exceeds in-state customer demand. Power purchasers such as the Bonneville PowerAdministration and PacifiCorp use Wyoming wind generation for portfolio diversity and sales to otherNorthwestern utilities, some of which are required to meet Renewable Portfolio Standards (RPS).1

Purchases of power generated in Wyoming to supply out-of-state demand suggest the need fortransmission line capacity development in conjunction with Wyoming generator development. However,factors such as separate ownership of generators and transmission lines, and high costs of transmissioncapacity additions, can temporarily decouple these developments. Transmission line congestion occurswhen scheduled power transmissions approach and exceed the established safety limits of transmissionline capacity. During periods of congestion, non-firm power generation, such as wind, is the first to becurtailed. In the case of the wind farm analyzed in this project, Foote Creek Rim I, transmission linecongestion was identified as a factor that could increasingly limit dispatch. Thus, a key focus of thisstudy was to compare the economic viability of an energy storage system at Foote Creek Rim I wind farmunder scenarios with and without transmission line constraints.

Another objective of this study was to incorporate, to the greatest extent possible, measured variation infactors affecting wind farm power deliveries. Projections of power output and revenues from wind farmoperations are typically based on average annual or monthly wind speed data, and do not includesimulated shorter term variation. Use of datasets with measured hourly wind speed and power generation,such as applied in this study, provide a more accurate view of wind farm operations without overstatingthe potential benefits of energy storage. The application of hourly transmission line load data to modelwind farm curtailment frequency and duration adds another layer of actual variation to this analysis.Together, these detailed datasets were used to conduct an economic assessment of a flow battery at FooteCreek Rim I that incorporated actual variations in wind speed, power generation, and transmission linecongestion.

1.2 The Site

Foote Creek Rim is a remote, treeless plateau in Carbon County, Wyoming. Foote Creek Rim is one ofthe windiest places in America, with extreme temperatures that fall as low as 30°F below zero in thewinter. The 41.4-MW Foote Creek Rim I wind farm is one of five wind farm operations on the plateau.There are 69 Mitsubishi 600-kW turbines installed along a three-mile portion of the plateau, with

1 RPS goals of 15 to 20% of statewide power to be provided by renewable resources by or before 2020 have been setin the western states of Nevada, California, and Washington.

2

elevation varying from 7,950 to 7,750 feet. This site began producing power in April 1999, and was thefirst commercial wind generated power station in Wyoming.

Foote Creek Rim I is owned by PacifiCorp2 and the Eugene Water and Electric Board (EWEB), and isoperated by AES SeaWest.3 The project pays substantial taxes to the county and state (i.e., $800,000/yearis received by the county).4 Purchasers of power from the wind farm include PacifiCorp, EWEB, and theBonneville Power Administration. Electrical facilities include a substation, and a 28.8-mile transmissionline connecting the facility to the grid.

1.3 Power Transmission and Congestion

In Wyoming, electricity is transmitted primarily from the northeast part of the state to the southwest alongPacifiCorp’s TOT 4A transmission line. A smaller capacity transmission line in the opposite direction isknown at TOT 4B. When the capacity of these lines limits the amount of energy that can be transferred tothe western load centers, the grid operator finds other (more expensive) generation sources on the loadside of the constraint. During these periods of constraint, or congestion, non-firm generators (e.g., windfarms) are among the first to be curtailed. In 2003, there were a total of 1,686 hours in which loadexceeded 75% of the capacity along TOT4, indicating congestion. During 187 of these hours, loadexceeded 90% of capacity.5

The strained Rocky Mountain regional system and its inability to accommodate new power generationhave been addressed by the Rocky Mountain Area Transmission Study (RMATS).6 Co-sponsored by theGovernors of Wyoming and Utah, RMATS was a consensus planning study conducted by regionalindustry, governmental, and environmental stakeholders in 2004 to examine transmission expansionneeds. RMATS recommendations included several projects to increase transmission capacity from inter-mountain states to more densely-populated surrounding states, including:

The Frontier Line, announced by the Governors of California, Nevada, Utah and Wyoming inApril 2005. This interstate high-voltage transmission line was proposed to originate in Wyomingwith terminal connections in Utah, Nevada and California.

The TOT3 Project, announced in September 2005 by the Wyoming Infrastructure Authority,Trans-Elect, and the Western Area Power Administration to strengthen the electrical tie betweenWyoming and Colorado.

TransWest Express, announced by Arizona Public Service in 2006, to reach Wyoming withtransmission lines from northern Arizona through Utah.

The Wyoming Infrastructure Authority and National Grid signed a Memorandum of Understanding(MOU) in December 2005 to jointly conduct the Wyoming – West study, which was to help lay thegroundwork for a significant increase in transmission capacity between Wyoming and neighboring statesto the West. The Wyoming - West study assessed the RMATS recommendations in light of subsequentproject announcements, and focused on identifying new transmission needs within Wyoming and betweenWyoming and its neighbors to the west. The MOU between National Grid and Wyoming InfrastructureAuthority addressed their continued relationship on permitting, financing, construction and operation of

2 PacifiCorp was acquired by Mid-American Holding Company in 2006.3 Previously SeaWest WindPower of San Diego, this company specializing in wind farm development and assetmanagement was acquired by AES in 2005.4 As reported by EWEB at http://www.eweb.org/Home/Windpower/wyoming/foote_creek.htm and by RenewableNorthwest Project at http://www.rnp.org/Projects/foote.html on 12/11/065 Further description of constraints along the TOT4 transmission line is provided in Appendix A.6 See http://psc.state.wy.us/htdocs/subregional/home.htm.

3

new transmission lines identified by the study. The time-frame in which new transmission projects comeon-line will undoubtedly affect the extent of future curtailment of intermittent power providers such aswind farms.

1.4 Flow Battery Contributions

A large multi-MWh energy storage system that can control energy discharge over hours (as opposed tominutes) could provide Foote Creek Rim I with storage of power generated during curtailment for latersale. Such an energy storage system could also increase the reliability of wind generation during periodsof peak demand, thereby enabling the wind farm to seek a conditional firm tariff. PacifiCorp is workingwith Bonneville Power Administration to develop a Conditional Firm product to be implemented by year-end 2008, enabling wind farms and other intermittent resources to receive the benefits of firm powerdelivery during pre-determined periods.7

The VRB Energy Storage System (VRB-ESS) can provide many hours of electrical energy storage in avanadium redox regenerative fuel cell (also known as a flow battery). Energy is stored in vanadium ionsin a dilute sulfuric acid electrolyte. The reaction is reversible, allowing many charge-discharge cyclesand deep discharges. The VRB-ESS can absorb or discharge energy within milliseconds in response toload fluctuations, providing voltage and frequency stabilization. As a result, the VRB-ESS can be used tostore power generated during periods of transmission line congestion (e.g., during curtailment), and torelease power in response to changes in dispatch orders.

PacifiCorp has been testing a VRB 250-kW/8-hour (or 2 MWh) flow battery at the Castle Valleysubstation in Moab, Utah. The Castle Valley system is the first flow battery storage system installed andoperated in the U.S. For this study, VRB conceptually scaled up the Castle Valley storage system, and re-oriented the electrolyte tanks to vertical to reduce the system’s footprint. The conceptual design appliedin this study is capable of either 6 or 8 hour discharge.

1.5 Project Objectives

The overall objective of this project was to examine the potential benefits of a flow battery installation atFoote Creek Rim I. The analytic approach was to:

Identify preferred battery charge and discharge periods based on periods of transmission linecongestion and wind variation at the wind farm.

Assess tariff rate variables on the economics of modeled flow batteries under scenarios with andwithout transmission line congestion.

Evaluate the effects of battery capital cost, salvage value, and rebate on battery systemeconomics.

Analyze the potential economic benefit of capturing spilled, light winds, enabled by theinstallation of an energy storage system.

7 On May 18, 2006, FERC issued a notice of proposed rulemaking to update Order 888, the landmark opentransmission access order that FERC issued in 1996. Transmission owners/operators are investigating conditionalfirm products and will likely file those tariffs with FERC. PacifiCorp, since its merger with MidAmerican EnergyHoldings, is pursing novel Conditional Firm products, with public stakeholder participation. Three products arebeing considered. Benefits to firm providers include capacity payments (in addition to energy payments) forsuccessful delivery of pre-determined amounts of power; and high priority dispatch (i.e., reduced chances ofcurtailment during congestion) throughout designated firm periods.

4

2 Methods

Excel spreadsheet models were developed to project wind farm power output and economics of an energystorage system under different tariff rates, with and without transmission line congestion. Datasets wereused to determine battery charge and discharge cycles and assess battery economics under differentconditions.

2.1 Datasets

The datasets used in this project, their sources, conversions, and applications were provided by teamparticipants after non-disclosure agreements were signed.

Wind Speed Data. Wind speed measurements (meters/second, m/s, in 10-minute intervals for the year2003) were provided by AES SeaWest under a non-disclosure agreement for six meteorologicalstations and each of the 69 wind turbines at Foote Creek Rim I. This dataset had to be reduced beforeit could be used in the analysis. An Excel model was used to reduce the 10-minute interval data tohourly averages and compare the wind speeds across the meteorological stations and turbines to provethat the wind speeds aligned. The wind speeds recorded at the six meteorological stations wereaveraged to create a single set of hourly wind speeds to represent the entire wind farm. This data wasused to assess wind speed variation over the year.

Turbine Power Generation. Power generation (watt-hours in 10-minute intervals over the year 2003)was provided by AES SeaWest under a non-disclosure agreement for each of the 69 wind turbines andsubstation at Foote Creek Rim I. Hourly power generation was calculated by adding six sequential 10-minute intervals. An Excel model was used to calculate power generation from the actual wind speedreadings at nine turbines located at discreet segments along the ridge. Wind speed and powergeneration data were validated. In addition, the power generation output from the substation wasvalidated against the summed generation output of all 69 turbines and found to be within tolerance(see Appendix B). Hourly power generation from the substation was used in the Excel model toproject hour-by-hour output from the wind farm.

Transmission Line Load. End-of-hour load data for TOT 4A and 4B were provided by PacifiCorpunder a non-disclosure agreement for the year 2003. This data was used to determine the periods ofgreatest congestion, and hence periods during which non-firm power from Foote Creek Rim I wasleast likely to be dispatched. Hourly load throughout the year was used to trigger curtailment in theExcel model.

VRB-ESS Energy Flow. One month of energy flow data from the 2-MWh VRB-ESS was provided byPacifiCorp. In Utah, the VRB-ESS was installed at the end of a long 25-kV feeder line strained bysignificant seasonal demand fluctuations – a very different application from that envisioned in thisstudy. The data from Utah was used in a preliminary assessment to confirm the round-trip efficiencyof the system and the fact that the VRB-ESS did not self-discharge while in an idle state.

2.2 Tariff Rates and Rebates

Large wind farms negotiate Power Purchase Agreements with the utilities buying their wind power.PacifiCorp is no longer vertically integrated, and as a result, different business units operate differentassets. The entity responsible for the wind farm was unwilling to share the terms of its Power PurchaseAgreement. PacifiCorp offers a conservative $0.035/kWh energy charge with no time-of-day or seasonalvariation and no capacity charge for small-scale purchase of renewable generation in Wyoming. This

5

investigation required a more aggressive tariff, with seasonal and time-of-day variation to encouragedischarge of the wind energy stored in order to maximize revenues. As a result, the analysis examinedtariffs used in California with capacity and energy charges and time-of-day factors as a proxy for thePower Purchase Agreement rates.

Tariffs were examined in two ways, as a “proxy” tariff based on an actual tariff, and as a simplified tariffto examine the economics of energy payments alone. The simplified tariff for examination of energy rateeffects had one season and two time-of-use (TOU) energy rates – a high weekday rate from noon to18:00, and a single low rate for all other times. The difference between high and low rates was modeledto vary from $0.05 to $0.15 to examine rate differences during battery charging and discharging requiredfor economic viability in absence of other cash inflows (i.e., capacity payments and rebates).

For all other analysis, a proxy tariff was used that was based on the generator portion of the unbundleddemand rates for a Pacific Gas & Electric (PG&E) tariff for medium commercial/industrial facilities (i.e.,Schedule E-19). This tariff designates two seasons, and three TOU periods (see Table 1). TOU periodswere rounded to whole hours for this project (i.e., 20:30 became 21:00).

Table 1. Seasonal TOU Periods Defined in Selected TariffSummer (May-Oct) Winter (Nov-Apr)

TOU PeriodStart Finish Start Finish

Day of week

Peak 12:00 18:00 M-F

Part-peak 9:00 12:00 9:00 22:00 M-F

18:00 22:00 M-F

Off-Peak 22:00 9:00 22:00 9:00 M-F

0:00 0:00 0:00 0:00 Sa-Su

Two types of payments are provided in this tariff – an energy payment and a demand payment. Thedemand payment is comparable to a capacity payment in a generator’s contract.8 The energy payment isdetermined from the amount of power delivery, and the tariff-designated Energy Rate. The capacitypayment is determined from the successful delivery of a pre-determined amount of power and the tariff-defined capacity price ($/kW). As shown in Table 2, the proxy energy rates and capacity rates vary byTOU period and season.

Table 2. Seasonal TOU Energy and Capacity Payments in the Proxy TariffEnergy Rates ($/kWh) Capacity Price ($/kW)

TOU PeriodSummer Winter Summer Winter

Peak 0.10334 NA 75 NA

Part-peak 0.07502 0.06613 NA NA

Off-peak 0.04903 0.0502 NA NA

Modeling a capacity payment assumes a conditional firm tariff, which would likely be enabled by anenergy storage system. Under the proxy tariff, the most lucrative period for selling power (i.e., battery

8 The PG&E Schedule E-19 demand payment was only $7.78/kW. It was thought that this demand payment is muchlower than typical capacity payments, thus we use a capacity payment of $75/kW.

6

discharge) is during the summer peak period, which has both a capacity payment and the highest energyrate. The best period for battery charging (i.e., reducing sales to the grid) is during the off-peak period,when energy rates are lowest and there is no capacity payment.

One final note regarding tariffs, the capacity payment carries a penalty for non-performance. This penaltyvaries by utility. The penalty rate used in this study was 1.25 times the capacity rate, which is the same asthat used in a recent California Energy Commission model.9 For the purposes of this study, the penalty isassumed to apply to each hour in which the delivered wind generation did not meet promised targets.

2.3 Excel Model for Storage System Analysis

An Excel-based model was developed for economic analyses of power generation and discharge from anenergy storage system. The model incorporates one year of hourly average data of wind speed and powergeneration data from Foote Creek Rim I, and requires input variables such as power storage systemspecifications and tariff prices (see Table 3).

Table 3. Energy Storage System Economic Model Input Variables

Storage System Parameters and Costs Tariff and Other Economic Variables

Round-trip efficiency losses (%) Minimum discharge level (%) Rate of charging and discharging (kW) Charge and discharge schedules (daily, weekly, and

seasonal) Battery capital costs Battery Operations & Maintenance (O&M) costs

($/kWh) Battery salvage value ($/kWh) and annual change in

value relative to constant dollars Expected life (up to 30 years) Capital depreciation (7, 10, 15, or 20 years)

TOU and season definitions Energy rates Capacity rates and penalty Minimum power delivery and percent chance

of failure to deliver during peak periods Transmission line load (%) at which

curtailment begins, and % curtailed at differentloads

Rebate amount and year of provision Federal and State tax rates (%) Rate of return

Based on this input, the model projects the quantity of power output to the grid during each hour of theyear for the wind farm (with no battery), and for the wind farm with an installed battery (see Table 4).Hourly power output is multiplied by the applicable tariff-defined payments. The difference betweenthese payments with and without a battery is calculated as the battery revenue. Battery costs are thensubtracted from this revenue to assess overall economic viability of the battery. Thus the economicanalysis in this study assesses the incremental change in cash inflow and outflow due to batteryinstallation and operation. For this study, the project life was defined as 20 years, with 15 years capitaldepreciation. All costs were developed in collaboration with the battery manufacturer.

When applicable, capacity payments were calculated as monthly capacity payments from the minimumexpected power delivery during the peak period, reduced by estimated monthly capacity penalties. Forthis study, minimum expected deliveries (or commitments) were estimated as the battery designeddischarge rate. Hourly generation data and modeled battery discharge were summed to determine whenpower output during the peak rate period failed to meet the hourly peak rate commitment.

9 Lamont, A., Improving the Value of Wind Energy Generation Through Back-up Generation and Energy Storage,California Energy Commission, CEC-500-2005-183, April 2004, p. 3.

7

Table 4. Energy Storage System Economic Model Output and Projection PeriodsProjection Period

OutputHourly Monthly Annual Life Cycle*

Battery discharge (kWh) ● ● ●

Output to grid ● ● ●

Energy revenue ● ● ●

Probability of capacity penalty (failure to meet commitment) ●

Capacity revenue (penalties incorporated) ● ● ●

Total revenue due to battery ● ● ●

Battery O&M costs ● ●

Rebate ● ●

Salvage value ●

Taxes (does not include possible taxes on rebates) ● ●

Tax reductions for other operations ** ● ●

Capital costs ●

Net Present Value (NPV) ●

Rate of Return when NPV = 0 ●* Life cycle cash flows are calculated as present value with the input rate of return.** Tax deductions for other operations are only applicable when tax deductions for capital depreciation exceed cash inflows.

2.4 Excel Model for Light Wind Capture

The cut-in and cut-out speeds for the Mitsubishi turbines at Foote Creek Rim I were set at 4.8 m/s and 27m/s, respectively (if temperatures were very cold, the turbines would cut out at lower speeds). This cut-inspeed is higher than many other similar turbines due to the high altitude of Foote Creek Rim, and relatedlow air density and wind power. Turbine cut-in speed is often set a little higher than the actual minimumwind speed at which turbine blades are moved and power is generated to reduce the number of times theturbine generator is turned on and off, and associated mechanical stress. Thus, while the Mitsubishiturbine controlling software allows easy re-setting of the cut-in speed, power gains from reducing the cut-in speed are not thought to be significant.10 In some cases, older turbines may be retrofit to allow lowercut-in wind speed, allowing the capture of currently spilled light winds.

In this section, the model for storage system analysis was modified to evaluate the potential benefits fromretrofits to capture some of the currently spilled light winds, recognizing this analysis is not likely to beapplicable to the Mitsubishi turbines at Foote Creek Rim I. Power generation from these light windswould not be sent directly to the grid, but could be directed to an energy storage system for conditioningand subsequent sale to the utility. Thus, the light wind capture retrofit is only possible after storagesystem installation, and could provide an additional benefit of the storage system. As in the model forstorage system analysis, the light wind model estimates the incremental costs and cash benefits of theretrofit. While battery size and tariff rates are incorporated in this analysis, they are only used to assessthe amount and value of captured light winds.

In contrast to the battery system cost estimates used in the storage system mode (i.e., developed incollaboration with the battery manufacturer), cost estimates for this enhancement are general estimates. Itwas estimated that the capital costs for implementing this retrofit could be $100,000. O&M costs

10 Personal communication with Mr. Utomi, Mitsubishi Engineering Manager for Foote Creek Rim I turbineinstallation, 4/3/07.

8

associated with the retrofit were estimated as 0.5% of the capital costs per year. Input variables used inthe model for light wind capture retrofit are shown in Table 5.

Table 5. Light Wind Model Input VariablesRetrofit Costs and Specifications

to Estimate Captured PowerStorage System Costs and

Specifications Tariff and Other Economic Variables

Retrofit capital costs Retrofit O&M costs Air density Rotor diameter Power coefficient for light

winds Current turbine cut-in speed Retrofit turbine cut-in speed Expected life (up to 10 years) Capital depreciation (3, 5, or 7

years)

Round-trip efficiency losses(%)

Minimum discharge level(%)

Rate of charging anddischarging (kW)

Charge and dischargeschedules (daily, weekly,and seasonal)

Battery O&M costs ($/kWh)

TOU and season definitions Energy rates Capacity rates and penalties Minimum power delivery and

percent chance of failure to deliverduring peak periods

Transmission line load (%) at whichcurtailment begins, and % curtailedat different loads

Federal and State tax rates (%) Rate of return

The model calculates and compares profits with and without spilled wind capture by calculatinghypothetical power generation for wind speed less than the current cut-in speed (4.8 m/s). The actualpower generation data included hourly generation during many hours with average wind speeds that wereless than the turbine cut-in speed. Because wind power increases to the cube of wind speed, smallincreases in wind speed cause a far greater change in wind power. Power generation data were adjustedas follows to create comparable datasets:

No Spilled Wind Capture: Power generation data with the current cut-in speed was altered bysetting all measured generation to zero during hours with an average wind speed < 4.8 m/s.

Spilled Wind Capture: Power generation during hours with an average wind speed < 4.8 m/s butmore than the retrofit cut-in was calculated based on the standard equation for wind power (i.e.,the product of half the air density, wind speed cubed, and rotor area) multiplied by the inputturbine and site-specific power coefficient for light winds.

All power generation from light winds was modeled to be sent to the battery (when it was not fullycharged) regardless of whether the battery was scheduled to be charging, discharging, or idling. In thisstudy, the power coefficient for light winds was input as the power coefficient for the lowest wind speedmeasured in the turbine manufacturer’s power curve (i.e., 13%). The economic benefits of the retrofitwere assessed based on the difference in profit with and without the retrofit, using an expected life of 10years, and capital depreciation over 5 years.

This method may overestimate the benefits of capturing light winds. First, setting all base-casegeneration to zero when the hourly average wind speed is less than the cut-in speed may overestimate thetime during which winds light winds may be caught with the retrofit. Second, setting the powercoefficient to be the same as for the lowest non-zero measurement in the turbine’s current power curve isoptimistic. At lower wind speeds, the power coefficient typically decreases with wind speed.

9

Frequency by Time of Day when Load Exceeds 75%Summer (May – October) Winter (November – April)

3 Time of Day Patterns in Data

The observed patterns of power generation, associated wind speed variation, transmission line congestion,and high-demand periods in tariffs were the primary factors considered in determining storage systemcycling schedules.

3.1 Transmission Line Congestion

In this study, wind farm curtailment was modeled to begin when the load on TOT4 reached 75% based onhourly averages. This provides a relatively high rate of curtailment to facilitate assessment of curtailmenteffects.11 As seen in Figure 1, congestion along PacifiCorp TOT4 primarily occurred in the early morninghours. According to PacifiCorp, these constraints were due to scheduled power flows to and from theNortheastern portion of the state in anticipation of major power plant outages for maintenance, and thusare not associated with periods of peak demand in the Western states. Hence, power transfer from thewind farm to the grid was most likely to be curtailed during early morning hours. This implies that thebattery should charge during the early morning and discharge at other times.

Another way to examine transmission line congestion is presented in Table 6 – counting the number ofhours during which transmission line load exceeded 75% and calculating the percentage of hours in eachtariff period that transmission was congested. Only 15% of the hours in the Summer witnessedtransmission line congestion whereas 23% of the Winter hours were congested. For Foote Creek Rim I,the hours within a day that were most likely to be subject to curtailment due to transmission linecongestion occurred during lowest tariff rate period. Transmission line congestion does not translate intototal curtailment of the wind farm. Rather, a small portion of the generation – 10% -- was modeled to becurtailed whenever the transmission line load exceeded 75%, increasing to 50% curtailed when thetransmission line load reached 100%. This approach was developed from expert opinion.

11 Actual curtailment of the wind farm due to transmission line congestion in 2003 is not provided in the availabledatasets. However, curtailment occurred substantially less frequently than indicated by the curtailment scenarioused in this analysis (i.e., beginning at 75% transmission line load).

Figure 1. TOT4 Transmission Congestion Profile

0

20

40

60

80

100

120

0:0

0

2:0

0

4:0

0

6:0

0

8:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

Time of Day (Hour)

Fre

que

ncy

.

0

20

40

60

80

100

1200:

00

2:0

0

4:0

0

6:0

0

8:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

Time of Day (Hour)

Freq

uen

cy.

10

Table 6. Transmission Line Congestion by Season and Tariff Period

Season Rate Period Total Hours Congested Hours % Time Congested

Peak 789 8 1%Part-Peak 920 21 2%Off-Peak 2,707 654 24%

SUMMER

Total 4,416 683 15%

Peak 0 0 0%

Part-Peak 1,681 245 15%Off-Peak 2,663 758 28%

WINTER

Total 4,344 1,003 23%

3.2 Wind Speed and PowerGeneration

The annual average wind speed at FooteCreek Rim I in 2003 was 10.65 m/s, andaverage generation was 14.7 MW (seeFigure 2). The 20th percentile of windspeed indicates that 20% of the time, windspeed was 4.48 m/s or less. The turbinesdid not generate power when wind speedfalls below 4.8 m/s. This suggests thatwind speeds at Foote Creek Rim I werevery low during one of every five hours.

Monthly average wind speed and powergeneration varied at Foote Creek Rim Ithroughout 2003 (see Figure 3). Averagemonthly wind speed during the summermonths (May through October) was 8.6m/s, with an average power generation of10.1 MW. During the winter months(November through April), averagemonthly wind speed was 12.7 m/s andgeneration 19.5 MW. Better turbineperformance was achieved in the winterdue to higher wind speeds, fewer periodsof light winds, and less frequent plannedmaintenance periods.

Figure 2. Power Generation and Wind SpeedPercentiles

AverageSpeed:

10.65 m/s

20thpercentile:4.48 m/s

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30

Wind Speed (m/s)

Figure 3. Monthly Average Wind Speed and PowerGeneration

0

2

4

6

8

10

12

14

16

18

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Win

dSpee

d(m

/s)

,

0

5

10

15

20

25Pow

erG

ener

atio

n(M

W)

,Wind Speed

Generation

11

In the summer there were morefrequent periods of light winds (speedsbelow 4.8 m/s). Figure 4 shows thenumber of consecutive hours withoutpower generation for each no-generation event in 2003. The longestevent was 29 hours in early November– substantially longer than thedischarge capability of the energystorage system modeled in this study.The density of the relatively short barsduring the summer months in Figure 4indicates that no-generation eventswere both shorter and more commonin the summer. Overall, during winterof 2003, there was no powergeneration for 11% of the total hours,and the average no-generation event lasted 4.9 hours. In contrast, there was no power generation for 17%of the total hours during the summer, and the average no-generation event lasted only 3.6 hours.

Hypothetically, if battery discharge is scheduled for six hours a day, the battery would charge for at leasteight hours a day for full recharge (compensating for round-trip efficiency losses). This suggests that no-generation events that last more than ten hours have the potential to prevent full battery recharge. Figure5 shows the frequency of no-generation events in the summer and winter by the number of consecutivehours with no power generation. In the summer, there were 18 no-generation events that exceeded 10hours in duration, while in the winter, there were only 13 no-generation events that exceeded 10 hours.As a result, it is not surprising that the battery was able to fully recharge on more days in the winter thanin the summer.

There was an inverse correlation between average hourly wind speed and the frequency of the no-generation events by TOU and season (see Figure 6). The most common time for no-generation eventswas between 6AM and 10AM in the summer, and between 2AM and 4AM in the winter. In both seasons,

Figure 4. Seasonality and Duration of No-GenerationEvents

Figure 5. No-Generation Events by Season, 2003

Winter (N o v . - A pr): 102 events , avg. 4.9 hr.Sum m er (M ay - Oct .): 219 events , avg. 3.6 hr.

0

5

10

15

20

25

30

J an F eb M ar A pr M ay J un J ul A ug Sep Oc t N o v D ec

WINTER: Nov. to Apr.Average Consecutive Hours with No Generation = 4.90

0

10

20

30

40

50

60

70

80

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29

Consecutive Hrs with No Generation .

Fre

quen

cy.

SUMMER: May to Oct.Average Consecutive Hours with No Generation = 3.63

0

10

20

30

40

50

60

70

80

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29

Consecutive Hrs with No Generation .

Freq

uen

cy.

12

average hourly wind speeds were greatest in the afternoon, and no-generation events were least commonin the mid-to-late afternoon.

The patterns of wind speed and corresponding power generation during 2003 suggest that the batterycould be charged more rapidly in the afternoon, when wind speeds were greatest. Particularly in thesummer, if battery charging is set during the lowest rate periods (i.e., morning hours), it would takelonger, on average, to fully recharge the battery, due to lighter winds.

3.3 Battery Cycling, Transmission Congestion, and Tariffs

In the model, the proxy tariff triggered battery cycling, charges during low rate periods and dischargesduring peak periods. For a wind farm with a non-firm tariff (i.e., with no firm delivery periods), batteryrecharge is preferred during periods with greatest congestion because curtailment would not affectcharging. At Foote Creek Rim I, peak congestion periods coincided with the lowest rate periods in theproxy tariff. As a result, there was no conflict in the optimal battery charge/discharge schedule based onthe combined consideration of congestion patterns and tariff rate periods.

Charging the battery during periods with the greatest wind speed and lowest frequency of no-generationevents can increase the probability of a fully charged battery at the onset of a “firm” period. In the case ofa conditional firm tariff with firm delivery hours set based on daily wind speed patterns at Foote CreekRim, the battery would charge during the afternoon, and discharge during other periods, despite greatertransmission line congestion. Congestion patterns can be given less consideration under a conditionalfirm tariff because during the designated “firm” period, the power provider is among the mostly likely tobe dispatched, and the least likely to be curtailed.

For the modeled scenarios in this study, battery cycling was scheduled to maximize the energy andcapacity charges that would accumulate from the proxy tariff rates and schedule. For economic viability,battery discharge periods require higher payments than battery charge periods. Payments during thedischarge period should be sufficiently high to compensate for round-trip efficiency losses during batterycycling. If the battery discharges and charges during the same rate period, there will be a loss of revenuedue to round-trip efficiency limitations. As a result, under the proxy tariff, there would be no benefitfrom battery discharge during weekends with continuously low rates.

With respect to weekdays, the energy rate at which the stored energy is sold needs to be at least 30%greater than the energy rate at which the battery is charged to compensate for battery cycling losses, plus

Figure 6. Wind Speed and No-Generation Events by TOU and Season, 2003

WINTER

0

10

20

30

40

50

60

0:00

2:00

4:00

6:00

8:00

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

Freq

uenc

yof

No

Gen

erat

ion

.

0

2

4

6

8

10

12

14

Ave

rage

Win

dSpe

ed(m

/s)

.

SUMMER

0

10

20

30

40

50

60

0:0

0

2:0

0

4:0

0

6:0

0

8:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

Fre

quen

cyof

No

Gene

ratio

n.

0

2

4

6

8

10

12

14

Ave

rage

Win

dS

peed

(m/s

).

13

the O&M costs (i.e., $0.008/kWh discharge). Using this rule of thumb, under the proxy tariff, energystored during an off-peak period can be sold either during the part-peak or peak rate periods, but chargingshould not occur during the part-peak or peak rate periods except during curtailment. Charging thebattery with curtailed power is essentially the same as charging at a payment rate equivalent to $0.

Depending on wind speeds, the battery charge period may need to be substantially greater than thedischarge period. In the model, the time needed for full battery recharge varied from 8 hours to more than24 hours. The battery charge and discharge schedule for each season are inputs to the model, and thereare no penalties for extended charge periods. Thus, the battery charge period was set to coincide with alloff-peak hours with the exception of one idle hour during the early morning on weekends. Duringsummer weekdays, the battery was idle during the part-peak period, and discharged during the peakperiod (i.e., weekdays, noon to 18:00). In scenarios with battery operation during the winter, the samedischarge period was applied, as there were no peak hours during the winter (see Tables 1 and 2 on page5). A battery recharge with curtailed energy was permitted whenever the battery was not fully charged.

14

4 Flow Battery Assessment

The battery economic model was exercised under a variety of conditions to assess the effects of keyvariables on revenue. Some findings are unique, such as a choice to not operate the battery duringseasons when tariffs are restrictive yet still achieve considerable annual income. Shortcomings of tariffsin place become evident from the discussion that follows.

4.1 Battery Discharge and Revenue

With the inputs described in the study, the model yielded unique discharge and revenue output for the 12-MWh flow battery proposed for Foote Creek Rim I (see Figure 7). Without transmission constraints, bothsummer and winter six-hour discharge matched the summer six-hour peak period of the proxy tariff, andthe battery was charged during the off-peak rate period. While battery discharge varied from month tomonth, this variation did not strongly correlate with season in the model. Battery revenue, however,varied greatly with season, with positive revenue in the summer months, and negative revenue (i.e.,losses) in the winter months. The negative winter revenue was a result of the small difference in thewinter energy rate during the battery charge and discharge periods (i.e., $0.016), compared to a $0.054difference during the summer. Note that the winter month with the lowest battery discharge (February)was the winter month with the smallest loss. This suggests that under conditions of no transmissionconstraints or other economic incentives, the battery should not discharge during the winter.

The model revenue curve for the battery was substantially improved under conditions of transmission linecongestion that resulted in wind farm curtailment (see Figure 8). Note that in months with little to nocurtailment (April and May, evident in Figure 8), curtailment scenario revenue is identical to the No-Curtailment revenue. The increase in revenue due to curtailment is generally greater during wintermonths because average wind speed is greater and more curtailed energy can be sent to the battery. Inthis scenario (i.e., based on TOT4 load data with 10% curtailment beginning at 75% load and increasinglinearly to 50% curtailment at 100% load), annual energy revenue due to the battery increased by a factorof 2.5. Revenue increases due to curtailment are particularly substantial because the effective energy rateof curtailed energy is $0. Thus, the rate differences between the periods of battery charge and dischargeduring the peak rate period are greater with curtailment than without curtailment (see Section 4.4 forfurther discussion of energy rate differentials).

Figure 7. Discharge and Revenue Modeled fora 12-MWh Battery

Figure 8. Modeled Revenue With and WithoutCurtailment for a 12-MWh Battery

0

50

100

150

200

250

300

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Dis

char

ge(M

Wh)

.

-4

-2

0

2

4

6

8

Rev

enue

($th

ousa

nds)

.

Battery Discharge

Battery Revenue

-3

0

3

6

9

12

15

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Rev

enue

($,T

hous

ands

). NoCurtailment

Curtail at 75%Load

15

4.2 Cash Inflow and Outflow

To assess the economic viability of a storage system, revenue and other cash inflow should be consideredin relation to cash outflow over the life of the project. Cash outflow includes the initial cost of the battery,O&M costs, and taxes. The battery manufacturer estimated annual O&M costs to be $0.008/kWh. Insome of the modeled scenarios, cash inflows were less than tax deductions for capital costs over a 15-yeardepreciation schedule.12 Thus, it was assumed that the modeled storage system would be part of a largerbusiness operation with substantial profit and taxes, from which the battery depreciation could bededucted. With this assumption, battery depreciation deductions applied to other operations yieldedreductions, or savings in taxes paid on these operations. Thus, profit would be calculated as the sum ofbattery revenue, rebates, and reductions in taxes paid on other operations when applicable, minus the sumof battery capital costs, O&M costs, and taxes paid on battery operations.

From an economic standpoint, the value of a payment (i.e., profit) today is greater than the value of thesame (nominal) payment in the future. This is due to the time value of money, which is based on theability to invest and earn interest on a payment received today, in addition to the risk that the futurepayment may not occur. The most common method for determining the present value of future cashflows adjusted for the time value of money is based on equations for the calculation of compound interest.When calculating present value of a future payment, the interest rate applied in these equations becomesthe discount rate. The discount rate is the same as the rate of return on investment, also referred to as theinternal rate of return (IRR).

The present value of cash inflow and outflow for each year of the project can be calculated and summedover the project life to assess overall economic viability. For assessments over the project life, a finalcash inflow to consider is battery salvage value. For the energy storage system modeled in this study, thecurrent salvage value of the battery vanadium is estimated to be $150/kWh. Over the past decade,vanadium spot market prices have increased on average, about 6% per year. For this analysis, weassumed that salvage values would increase on average, 3% per year over a 20-year project life. Usingthese assumptions, the 20-year present value of salvage for the battery modeled would be $483,240,assuming a 10% rate of return.

Figure 9 presents life cycle (20-year) present values of cash inflow and outflow for the 12-MWh batterywith no transmission line constraints. A 10% rate of return was applied in this analysis. Scenarios A, B,and C in Figure 9 represent increasing sources of cash inflow:

Energy payments only Energy payments and a capacity payment Energy payments with a capacity payment and a rebate

The battery was modeled to run only during the summer months, when the energy payment was greatest,and when the successful negotiation of a capacity payment was most likely. (As seen by the winter lossesin Figure 7, winter operation of the battery was not economically viable under the PG&E tariff -- thisextends to scenarios with a capacity payment that is only available during the summer.) The modeledcapacity payment was $75/kW per summer month. For capacity payment calculations, the power deliverycommitment during the six-hour peak summer rate period was 2 MW per hour with capacity penalties of1.25 times the payment goal when the battery was unable to achieve this delivery. The rebate modeled

12 Based on MARCS 15-year depreciation rates using the half-year convention for bringing equipment into service(i.e., the depreciation tables used by the IRS.)

16

was $800 per kW capacity paid once, at the end of the second year of operation. This rebate was withinthe range of the rebates anticipated in California. It was assumed that no taxes would be paid on therebate, and that the rebate would be applicable with battery operation limited to the summer.

Capital costs, O&M costs, salvage value, and energy payments shown in Figure 9 are the same for allscenarios. When life cycle cash inflow equals or exceeds cash outflow, the battery operation iseconomically viable with a 10% rate of return. While energy payments may be greater under other tariffs,additional forms of revenue are needed for a system with capital costs such as those of flow batteries.Depending on the required rate of return for the project, additional cash inflow in the form of a rebate, oroutflow reduction in the form of a subsidy might be needed for economic viability.

Overall, the economical feasibility of the modeled battery required greater cash inflows than could beobtained from energy payments alone. A combination of favorable capacity payments and rebates orsubsidies might enable economic viability.

4.3 Effects of Curtailment

Under the transmission congestion conditions modeled, curtailment does not affect capacity payments. Inthis analysis, capacity payments are only affected by battery discharge and wind power generation duringthe peak summer period. Summer curtailment occurs during the off-peak period, when the battery ischarging. During the charging period, power is preferentially directed to the battery up to the maximumcharging rate, thus curtailment during charge does not affect battery cycling. (It should be noted thatcurtailment may affect capacity payments at other sites served by transmission lines that have congestionduring the peak periods.) Likewise, a rebate would not be affected by the curtailment modeled in thisanalysis.

Figure 9. Cash Inflow and Outflow for a 12-MWh Battery Under Three Revenue Scenarios

Note: 20-year present value, 10% IRR.

$ (millions)

17

Due to the transmission load patterns in Wyoming, the primary cash flow affected by curtailment in thismodel was energy payments. The primary cash flows affected by summer-only versus full-year batteryoperation are energy payments and battery O&M and a small effect on taxes. Figure 10 shows the 20-year present value of energy payments and O&M costs under conditions of full-year and summer-onlyoperation, with and without curtailment.

Under conditions of no transmission constraints, revenue was greater and O&M costs were least when thebattery was modeled to operate only during the summer months. In contrast, under the curtailmentscenario, revenue was greater with full-year battery operation. While O&M costs also increased withfull-year operation, this increase was less than the increase in energy revenue.

Overall, transmission line congestion increases energy revenue when it occurs primarily during off-peakperiods. Furthermore, basic operational decisions, such as which months to operate the battery, can besubstantially affected by the extent of wind farm curtailment.

4.4 Battery Size and Energy Rate Differentials

The sum of the present value of cash outflow and inflow is referred to as the net present value (NPV). AnNPV equal to zero (0) or a positive number suggests cash inflow equal or exceed outflows, therebyindicating the project is economically viable under the rate of return applied. An acceptable rate of returnvaries with the interest rates of loans required for capital purchases, and on the investor’s requirements forreturn on their investment.

Recognizing that the energy rates in the proxy tariff are not sufficient for economic viability, a simplifiedanalysis of the effects of energy rate differences between high and low rates (i.e., corresponding to batterydischarge and charge periods) was conducted. This analysis used a single 6-hour high rate period(weekdays, noon to 18:00) and 18-hour low rate period with no seasonal differences. Energy ratedifferentials of $0.05, $0.10, and $0.15 were modeled for a series of batteries designed for 6-hourdischarge with capacity ranging from 1 to 6 MW, with resulting energy ratings ranging from 6 MWh to

Figure 10. Energy Revenue and O&M Costs by Operation and Curtailment

18

36 MWh. The NPV was set to 0 by changing the rate of return. The rate of return can then be comparedto a company’s or investor’s requirement.

As seen in Figure 11, neither $0.05 nor $0.10 differentials between low rate and high rate periods yieldeda positive rate of return based on revenue solely from energy payments. Regardless of the ratedifferential, the rates of return in Figure 11 increase under the scenario with wind farm curtailment (i.e.,beginning when TOT4 load reaches 75%). A $0.15 differential between energy rates during batterycharge and discharge periods provides a positive rate of return for all battery sizes modeled.

Of the battery sizes shown in Figure 11, the 2-MW battery yielded a slightly higher rate of return (hence a12-MWh battery with 2-MW capacity and a 6-hour discharge period was modeled in Figures 7 though11). The benefits of the 2-MW battery are easiest to see in the $0.15 differential curves, when the 2-MWbattery has a 0.85% rate of return with no curtailment, and a 1.25% rate of return with curtailment. Incontrast, the next highest rates of return are for the 4-MW battery, which are 0.81% and 1.01% under nocurtailment and curtailment scenarios, respectively. It should be noted that for most investors, a rate ofreturn less than 5% is not likely to be viewed favorably. Thus, the analysis shown in Figure 11 suggeststhat capacity payments, and/or rebates will be needed for economic viability of the modeled batterysystem, regardless of battery size.

Figure 11. Effect of Energy Rate Differential on Rate of Return

19

5 Capture of Light Winds

One of the premises of this study was that the installation of a flow battery could capture and conditionspilled winds (below the turbine cut-in speed) for an additional source of battery recharge. It is notthought that the ability to capture light winds can be further enhanced in the installed turbines at FooteCreek Rim. However, there may be some wind power generators that could be improved to capture andgenerate electricity from light winds. This analysis provides an initial assessment of the potential value ofimproved capture of light winds.

5.1 Output to Grid and Revenue

Under the modeled enhancements to capture spilled, light winds was assessed with a 12-MWh batteryunder the proxy tariff. For this analysis, the new cut-in wind speed was 2.5 m/s, and a 13% powercoefficient used to calculate power generation from wind speed between 2.5 and 4.8 m/s (the original cut-in speed).

Figure 12 shows monthly increases in power output to the grid and associated energy revenue due to thecapture of spilled winds. Compared to power output to the grid, energy revenue is proportionately greaterin the summer, when energy rates are greater. Both curves in Figure 12 are roughly the inverse of themonthly average wind speed curve shown in Figure 3 (in Section 3.2).

As shown in Figure 13, capacity revenue (only available during summer months) is also increased by thecapture of spilled, light winds. The increase in capacity revenue is due to a decrease in capacity penaltiesas a result of a more fully charged battery, and the resulting decreased chance of failing to meet deliverycommitments during the peak period. Although this reduction in the chance of penalties is only a fewpercent and applies to only one season, it causes greater capacity revenue increases than seen in energyrevenues throughout the full year. Reduced chances of failing to meet peak delivery commitments mayalso improve the ability to successfully negotiate a conditional firm tariff.

Figure 12. Energy Revenue and Output to GridIncreases from Spilled Wind

Figure 13. Capacity Revenue Increase andFewer Penalties from Spilled Wind

ChanceofPenalty

CapacityRevenue

0%

1%

2%

3%

4%

5%

May Jun Jul Aug Sep Oct

Red

uctio

nin

Pen

alty

Cha

nce

.

0

1

2

3

4

5

6

7

8

9

Rev

enue

Incr

ease

($Tho

usan

ds)

.

20

5.2 Cash Inflow and Outflow Comparisons

Life cycle (10-year present values) cash inflow and outflow for the light wind capture enhancement to themodel are displayed in Figure 14. A 10% rate of return is applied for this analysis. The scenarios havedifferent cash inflows, with Scenario A having only energy revenue (using the proxy tariff), and ScenarioB having energy revenue in addition to a $75/kW/summer month capacity payment. Scenario C is thesame as Scenario B, except the battery (and hence the light wind capture) is only operated during thesummer months, when the capacity payment is available. While it is beneficial from the standpoint ofenergy revenue to capture light winds throughout the year, the analysis of battery economics suggests itmay be preferable to operate the battery during periods when a capacity payment is available (particularlywhen curtailment is minimal).

Figure 14. Cash Inflow and Outflow for Spilled Wind Capture Under Three Revenue Scenarios

Note: 12-MWh battery, 10-year present value, 10% IRR.

The greater cash inflow compared to outflow for Scenarios B and C suggest potential economic viabilityof light wind capture when there is a capacity payment and associated penalties for not meetingcommitments. In contrast, greater cash outflows than inflows in Scenario A suggest that the spilled, lightwind capture is not economically viable when the revenue is only from the energy payments availableunder the proxy tariff.

5.3 Effects of Curtailment

The transmission line congestion and associated curtailment modeled in the study did not affect capacitypayments due to the capture of light winds (i.e., curtailment does not change the chances of penalties).Likewise, modeled curtailment does not affect retrofit capital costs or O&M costs (which were estimatedbased solely on capital costs). However, life cycle (10-year present value) energy revenue due to thecapture of spilled, light winds is slightly reduced by curtailment, as shown in Figure 15. There is also asmall effect on taxes, not shown. The negative effect of curtailment on light wind capture economics is aresult of the reduced availability of storage capacity for power generation from light winds. In otherwords, light wind power competes with curtailed power for storage capacity.

21

Figure 15 shows the change in energy revenue with and without curtailment for both full-year batteryoperation, and summer-only operation. The reductions due to curtailment were proportionately quitesimilar (i.e., curtailment caused a 6% energy revenue reduction in the summer, and a 5% reduction overthe full year).

Figure 15. Effect on Curtailment on Spilled Wind Energy Revenue

Note: 10-year present value.

22

6 Conclusions

This study has found that a large multi-MWh energy storage system could provide Foote Creek Rim Iwind farm with storage of power generated during periods of transmission congestion for later sale. Suchan energy storage system could also increase the reliability of wind generation during periods of peakdemand in other markets, thereby enabling the wind farm to seek a more beneficial tariff, such as theconditional firm tariff.

Under the scenarios examined in this study, with key input parameters including flow battery capital costsand battery efficiency, key findings include:

Energy payments alone may be sufficient to provide a positive return on flow battery investmentif the difference between charge and discharge energy rates exceeds $0.15.

Capacity payments are needed when there are no substantial subsidies and the energy ratedifferential between battery charging and discharging is less than $0.15 (which is usually thecase).

The absence of capacity payments during a portion of the year may make it financially preferableto not operate the storage system during these seasons.

The ability to meet power delivery commitments during specified periods is substantiallyincreased by an energy storage system, which may facilitate successful negotiation of aconditional firm tariff with capacity payments.

Financial viability of an energy storage system may be augmented by factors such as taxincentives, rebates and other subsidies, and high salvage values.

Capture of spilled light winds can increase battery state of charge and resulting revenue fromdischarges to allow an acceptable return on a modest (i.e., $100,000) investment to retrofit thewind farm.

By increasing battery state of charge, the capture of spilled light winds limits the ability of thebattery to store curtailed wind generation during transmission congestion periods.

Under transmission line congestion scenarios that result in curtailment, the ability to store energyfor later sale can significantly increase revenue.

Curtailment has the potential to make it financially preferable to operate an energy storage systemwhen it would otherwise be uneconomical.

23

Appendix A. Transmission Line Constraints

The Northwest transmission grid is used in ways that weren’t envisioned when it was built. Generationand transmission patterns have changed dramatically and the system is showing considerable stress. Amajor consequence is the growing number of times when the system is congested and operating outside ofits operating transfer capability (the industry standard for reliability). Demand on portions of theNorthwest grid is particularly high in the summer when large amounts of power are moving intoCalifornia. The Western Electricity Coordinating Council (WECC) assesses penalties on transmissionowners whenever the operatingtransfer capability is exceededlonger than 20 or 30 minutes.

Electricity is transmitted inWyoming primarily from theNortheast to the Southwest alongTOT 4A. A smaller capacitytransmission path in the oppositedirection is known at TOT 4B.TOT 4A and 4B are not assignificantly constrained as thelines they continue on, e.g., JimBridger West, which is THE mostheavily utilized transmission pathin the West (see Figure A-1).

The system load curve forTOT4 was calculated fromhourly load data for TOT4Aand TOT4B and a WECCreport provided byPacifiCorp (see Figure A-2).

Figure A-1. Transmission Paths Impacting the Northwest

Source: PacifiCorp, Conditional Firm Public Stakeholder Meeting, Oct. 23, 2006,http://www.pacificorp.com/File/File69023.ppt

Figure A-2. Calculating TOT4 Transmission Line Loads

Equation for the System Load CurveTOT 4B (MW) = X - 8.113 * EXP(0.0051* TOT 4A (MW))

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500 600 700 800 900

TOT 4A Load (MW)

TO

T4B

Load

(MW

)

24

This equation enabled an examination of the daily system load on TOT4 during 2003. Two weeks arehighlighted in Figures A-3 and A-4: July 25-August 1 and December 14-21. The periods of transmissioncongestion are easily visible.

In 2003, there were 187 hours spread over 47 days in which load duration exceeded 90% (RED oval) and1,686 hours spread over 246 days in which load duration exceeded 75% (ORANGE dashed oval).

Figure A-3. TOT4 Daily System Load – Week in July 2003

0

20

40

60

80

100

7/25 7/26 7/27 7/28 7/29 7/30 7/31 8/1

Sys

tem

Load

(%)

Figure A-4. TOT4 Daily System Load – Week in December 2003

0

20

40

60

80

100

12/14 12/15 12/16 12/17 12/18 12/19 12/20 12/21

Sys

tem

Load

(%)

Figure A-5. Calculating TOT4 System Load Duration

0

20

40

60

80

100

0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000

Annual Hours

Load

(%)

25

Appendix B. Wind Speed and Power Generation

Foote Creek Rim I runs almost three miles atop a ridge with elevation varying from 7,950 feet down to7,750 feet (see Figure B-1). SAIC built and tested Excel models to validate wind speed measurementsacross the 69 wind turbines and six meteorological stations and wind generation output from actual windspeeds and turbine operating algorithms. Significant seasonal variations were found, but wind speedmeasurements across the 69 turbines and six meteorological stations were found to be sufficiently similarto permit use of wind generation at the substation as the input into a model (see Figure B-2).

Winter is the season with the highest wind speeds in Wyoming, averaging 12.7 meters/second. Windgeneration in the winter is nearly double the summer average of 10.16 MW per hour. Winter windgeneration averages 19.5 MW per hour.

Figure B-1. Foote Creek Rim I Layout

Source: SeaWest, December 2004.

26

The Mitsubishi turbines used in Foote Creek Rim do not generate power at wind speeds below 4.8meter/second or above 27 meters/second (if temperatures are very cold, the turbines will cut off at lowerspeeds). These cut-in and cut-out speeds are shown as red lines in Figures B-3 and B-4. In Figure B-3,the blue line at the bottom of the graph indicates that turbine #44 was out of service (probably forscheduled maintenance) during the week of December 14-20, 2003. As shown in Figure B-4, summerwind speeds never reached the 27 meters/second cut-off. On average, summer wind speeds were 8.6meters/second. During this week, turbine #66 was out of service, as seen by the aqua line at the bottom ofFigure B-4.

Figure B-2. Errors in Wyoming Wind Speed Measurements

Source: Calculated from SeaWest Data on Wind Speeds at Turbines and Meteorological Stations, December 2004. .

Figure B-3. Wind Speed and Turbine Output in Winter

Source: Calculated from SeaWest Data on Wind Speeds at Turbines and Meteorological Stations, December 2004.

27

Figure B-4. Wind Speed and Turbine Output in Summer

Source: Calculated from SeaWest Data on Wind Speeds at Turbines and Meteorological Stations, December 2004.

28

Appendix C. Model Snapshots

The following three screen snapshots illustrate how the economic model is structured, with Input, Output,and Data & Calculations worksheets.

Figure C-1. Wind Battery Economic Model Input Page

29

Figure C-2. Wind Battery Economic Model Output Page

30

Figure C-3. Wind Battery Economic Model Data and Calculation Page